發布日期2026/02
推論速度124 tokens/s
首字延遲時間15.06s
輸入價格$2/1M
輸出價格$12/1M
排名表現
評測分析
Gemini 3.1 Pro Preview is well-suited for developers, researchers, and analysts who require strong reasoning and solid coding skills. However, avoid using it for complex image processing or autonomous agent workflows; pair it with a specialized multimodal model if vision is required.
強項
- Strong logical reasoning with top-tier benchmark rankings, well-suited for complex problem-solving and analysis
- Solid coding capabilities with competitive code generation and review, fitting software engineering workflows
- Natural conversational flow and strong language comprehension, suitable for drafting content, brainstorming, and Q&A
弱項
- Very weak multimodal capabilities (image, video, OCR), ranking #33 and making it unsuitable for vision-heavy applications
- Unstable autonomous agent performance (#41), making it unreliable for highly autonomous workflows
適用情境
Reasoning and analysis: Solving logic problems, qualitative data analysis, and researchProgramming: Code generation, refactoring, debugging, test writing, and technical reviewsDrafting and ideation: Brainstorming, article writing, planning, and creative dialogue
指南與影片
Visit gemini.google.com or the Gemini mobile app to use Gemini 3.1 Pro Preview. The model is tailored for complex tasks like programming, data analysis, video processing, and abstract reasoning. Tip: Google AI Pro subscribers ($19.99/month) get full access, or use Google AI Studio for free tier access with rate limits.